Workstream Continuity Design: Design Bible

Workstream Continuity Design (WCD) is a proposed HCI and product-design discipline for software in which people must repeatedly enter, leave, understand, supervise, and safely resume concurrent work carried by humans, services, and AI agents. It treats every focus transition—from seconds to days—as an orientation event and asks whether the system can reconstruct the minimum sufficient operating picture: goal, attention claim, operational state, meaningful change, responsibility, authority, evidence, consequence, and the safest useful next action—without requiring the operator to replay a transcript or rebuild context from raw activity.

This research-edition design bible defines workstream continuity as a product-level quality attribute and Workstream Continuity Design as the coordinating practice for designing that attribute. It develops a five-commitment framework; category boundaries and research lineage; a nine-slot continuity grammar and operational-diff vocabulary; a canonical information architecture; durable workstream, actor, agency, and control-posture models; acquisition and re-entry protocols; an attention and prioritization model; a 30-pattern library; visual, accessibility, human-oversight, and safety guidance; a 31-item anti-pattern catalog; metrics, experiments, and conformance tests; a maturity model; and an applied AI-first CRM case study.

The report also proposes a modular WCD Semantics Policy for the accountable surface of operated AI systems. Its core Accountable Expression Profile requires consequential machine expressions to be typed, attributable, supported by evidence and provenance, bounded in consequence, resolved against authoritative policy, linked to durable workstreams, and paired with proportionate human-intervention and recovery paths. Versioned extension profiles address instruction and context provenance, multi-principal identity and delegation, memory mutation, recurrence and steering, tool trust boundaries, multi-agent composition, configuration identity, resource authority, conformance evidence, and semantic regression testing. WCD is positioned as an application and conformance layer over existing agent, event, provenance, observability, state-synchronization, and declarative-interface standards—not as a replacement transport protocol.

Epistemic status: the underlying human-factors, interruption, situation-awareness, supervisory-control, and trust-calibration premises are grounded in established research. The category terminology, continuity grammar, architecture, semantics-policy profile suite, pattern library, metrics, maturity model, and conformance methods are original synthesis and proposals requiring implementation, cross-domain evaluation, and field validation. This non-peer-reviewed research edition is intended for product designers, HCI researchers, agent-platform teams, enterprise UX architects, safety and governance teams, and engineers building operational AI systems.